The Need for Ethnoracial Equity in Artificial Intelligence for Diabetes Management: Review and Recommendations

 
CONTINUE READING
JOURNAL OF MEDICAL INTERNET RESEARCH                                                                                                           Pham et al

     Viewpoint

     The Need for Ethnoracial Equity in Artificial Intelligence for
     Diabetes Management: Review and Recommendations

     Quynh Pham1,2, PhD; Anissa Gamble1, MSc; Jason Hearn1,3, MHSc; Joseph A Cafazzo1,2,4, PEng, PhD
     1
      Centre for Global eHealth Innovation, Techna Institute, University Health Network, Toronto, ON, Canada
     2
      Institute of Health Policy, Management and Evaluation, Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada
     3
      Faculty of Medicine, Memorial University of Newfoundland, St. John's, NL, Canada
     4
      Institute of Biomedical Engineering, University of Toronto, Toronto, ON, Canada

     Corresponding Author:
     Quynh Pham, PhD
     Institute of Health Policy, Management and Evaluation
     Dalla Lana School of Public Health
     University of Toronto
     Health Sciences Building
     155 College Street
     Toronto, ON, M5T 1P8
     Canada
     Phone: 1 4163404800 ext 4765
     Email: q.pham@uhn.ca

     Abstract
     There is clear evidence to suggest that diabetes does not affect all populations equally. Among adults living with diabetes, those
     from ethnoracial minority communities—foreign-born, immigrant, refugee, and culturally marginalized—are at increased risk
     of poor health outcomes. Artificial intelligence (AI) is actively being researched as a means of improving diabetes management
     and care; however, several factors may predispose AI to ethnoracial bias. To better understand whether diabetes AI interventions
     are being designed in an ethnoracially equitable manner, we conducted a secondary analysis of 141 articles included in a 2018
     review by Contreras and Vehi entitled “Artificial Intelligence for Diabetes Management and Decision Support: Literature Review.”
     Two members of our research team independently reviewed each article and selected those reporting ethnoracial data for further
     analysis. Only 10 articles (7.1%) were ultimately selected for secondary analysis in our case study. Of the 131 excluded articles,
     118 (90.1%) failed to mention participants’ ethnic or racial backgrounds. The included articles reported ethnoracial data under
     various categories, including race (n=6), ethnicity (n=2), race/ethnicity (n=3), and percentage of Caucasian participants (n=1).
     Among articles specifically reporting race, the average distribution was 69.5% White, 17.1% Black, and 3.7% Asian. Only 2
     articles reported inclusion of Native American participants. Given the clear ethnic and racial differences in diabetes biomarkers,
     prevalence, and outcomes, the inclusion of ethnoracial training data is likely to improve the accuracy of predictive models. Such
     considerations are imperative in AI-based tools, which are predisposed to negative biases due to their black-box nature and
     proneness to distributional shift. Based on our findings, we propose a short questionnaire to assess ethnoracial equity in research
     describing AI-based diabetes interventions. At this unprecedented time in history, AI can either mitigate or exacerbate disparities
     in health care. Future accounts of the infancy of diabetes AI must reflect our early and decisive action to confront ethnoracial
     inequities before they are coded into our systems and perpetuate the very biases we aim to eliminate. If we take deliberate and
     meaningful steps now toward training our algorithms to be ethnoracially inclusive, we can architect innovations in diabetes care
     that are bound by the diverse fabric of our society.

     (J Med Internet Res 2021;23(2):e22320) doi: 10.2196/22320

     KEYWORDS
     diabetes; artificial intelligence; digital health; ethnoracial equity; ethnicity; race

     http://www.jmir.org/2021/2/e22320/                                                                   J Med Internet Res 2021 | vol. 23 | iss. 2 | e22320 | p. 1
                                                                                                                        (page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH                                                                                                     Pham et al

                                                                           data represent the population that will ultimately be affected by
     Introduction                                                          the algorithm [42]. As health data have traditionally been
     There is clear evidence to suggest that diabetes does not affect      collected on predominantly White populations [43] or have
     all populations equally [1]. Among adults living with diabetes,       simply omitted relevant ethnoracial information [44], algorithms
     those from ethnoracial minority communities—foreign-born,             trained on such data are at risk of ignoring race and ethnicity.
     immigrant, refugee, and culturally marginalized [2]—are at            Such ethnoracial disparities have long been present in clinical
     increased risk of poor health outcomes [3-6]. Numerous studies        decision support tools, with various algorithms being arbitrarily
     have reported ethnoracial differences in glycemic control [7,8],      corrected for race with little or no scientific justification [45].
     diabetes prevalence [9], risk of diabetes complications [10], and     These algorithms are widely used to inform important clinical
     diabetes-related mortality [11]. Data from the Centers for            actions such as specialist referrals [46,47] and assess candidacy
     Disease Control indicate that non-Hispanic Black people are           for particular interventions [48,49]. In AI, where the effects of
     2.3 times more likely to die from diabetes than their                 biases may be dramatic and difficult to identify [50], careless
     non-Hispanic White counterparts [12]. Similarly, young people         incorporation of ethnoracial data may perpetuate health
     living with diabetes from Black or Hispanic backgrounds are           inequities for those communities in most need. The alarming
     at increased risk of poor long-term glycemic control when             potential for clinical decision support tools to be algorithmically
     compared to White youth [13]. The social determinants of health       biased in favor of advantaged populations demands careful
     describe the social, economic, and physical conditions in which       evaluation to promote their ethnoracial inclusivity. We believe
     people are “born, live, learn, work, play, worship, and age,” as      that to optimize equitability, AI research should (1) establish a
     well as the impact that such environments have on health              training population that is representative of the general
     outcomes [14]. As a result of the well-accepted contribution of       population, (2) report the ethnoracial distribution of the training
     the social determinants toward diabetes outcomes [15], we know        set, and (3) discuss potential ethnoracial limitations of the
     that ethnoracial minority populations are also more likely to         training data. To our knowledge, these simple tenets are not
     experience socioeconomic adversity and subsequent challenges          being met in existing diabetes AI research.
     with diabetes management and access to care [16]. This                As a digital health research group preparing to build AI-based
     association likely follows from the increased prevalence of           diabetes management tools [51], we want to address the
     various diabetes risk factors (eg, low birth weight, physical         challenges of promoting equity in AI and derive
     inactivity, obesity, smoking) in individuals of low                   recommendations that can inform our work and the field at
     socioeconomic status (SES) [16-18]. In Canada, where 21% of           large. In an effort to better understand whether diabetes AI
     the population are foreign-born and live in the nation’s largest      interventions are being designed in an ethnoracially equitable
     urban centers [2], people with a household income less than           manner, we conducted a rapid case study whereby we assessed
     Can $15,000 (US $11,745) are 4 times more likely than those           articles curated in an existing literature review of AI-based
     with a household income greater than Can $80,000 (US $62,635)         diabetes management tools. Our objectives were to (1) review
     to be diagnosed with type 2 diabetes (T2D) [19]. For people           ethnoracial considerations reported in past articles on AI-based
     living with type 1 diabetes (T1D), low SES has been associated        diabetes support tools and (2) propose a strategy to promote
     with an increased risk of poor glycemic control [20], as well as      ethnoracial equity in such tools in the future. This viewpoint
     higher levels of mortality and morbidity [21].                        serves to document the findings from our case study and the
     Innovative technologies are actively being researched and             recommendations proposed by our group to advance ethnoracial
     developed to mitigate the burden of diabetes on patients and          equity in diabetes AI.
     the health care system. Among the potential solutions, artificial
     intelligence (AI) appears to be well suited for diabetes              Case Study
     management given that this chronic condition has long been
     guided by quantitative data collected by patients, their devices,
                                                                           Methods
     and their care providers [22]. These data can be computationally      We conducted a secondary analysis of 141 articles included in
     complex for patients to make sense of on their own to inform          a high-impact literature review published in the Journal of
     their diabetes management [23]. AI, or the ability for “computers     Medical Internet Research in 2018 by Contreras and Vehi
     to think like humans” [24], has revolutionized many consumer          entitled “Artificial Intelligence for Diabetes Management and
     technologies (eg, facial recognition, fraud detection, self-driving   Decision Support: Literature Review” [52]. The selected review
     vehicles) and is now gaining momentum in the health care field.       included articles describing AI technologies for diabetes
     AI technologies are being developed for various areas of              management and decision support, as well as their associated
     medicine such as medical imaging analysis [25-27],                    challenges. We chose this review over comparable syntheses
     prognostication [28-30], and clinical decision support [31-33].       of the literature based on the short time since publication, the
     In diabetes care, AI is being applied for blood glucose prediction    breadth of diabetes AI interventions included for review, and
     and control [34,35], identification of adverse events [36,37],        the impact that the review has had on informing the diabetes
     lifestyle support [38,39], and predicting diabetes risk [40,41].      AI field.
     Despite the potential applications of AI in diabetes care, several    Two members of our research team independently reviewed
     factors may predispose these technologies to ethnoracial bias.        each of the 141 articles and selected those reporting ethnoracial
     The effectiveness of an AI algorithm is largely dependent on          data for further analysis. Articles were selected for analysis if
     the quality of training data, as well as how accurately training      they included an explicit description of participants’ ethnic

     http://www.jmir.org/2021/2/e22320/                                                             J Med Internet Res 2021 | vol. 23 | iss. 2 | e22320 | p. 2
                                                                                                                  (page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH                                                                                                    Pham et al

     background, racial background, or both. Articles were excluded
     if they were review papers, selected participants from a single
                                                                          Discussion
     ethnoracial group, or were inaccessible by the research team.        Ethnoracial Inequities in Diabetes AI
     The following criteria were charted for each of the selected
     studies: article type, diabetes type, ethnicity distribution, race   Diabetes AI programs are intended to improve diabetes-related
     distribution, number of participants, and source of data (ie,        health outcomes, experience, and expenditure [63,64]. However,
     electronic medical record, electronic health record).                it is unclear whether such systems benefit all populations
                                                                          equally. In our informal case study of 141 articles related to
     Results                                                              AI-based diabetes tools, we identified only 10 articles that
     In screening the 141 articles included in the Contreras and Vehi     specifically reported the ethnic or racial distribution of their
     review, only 10 (7.1%) were ultimately selected for secondary        studied patient population. We believe that this paucity of
     analysis in our case study [53-62]. Of the 131 excluded articles,    ethnoracial data in the reviewed articles significantly limits the
     118 (90.1%) failed to mention participants’ ethnic or racial         effectiveness of the associated AI technologies. Several
     backgrounds. The remaining articles were excluded because            examples of such ethnoracial bias in clinical algorithms have
     they were review papers (n=5), selected participants from a          been previously reported in the literature [42,45]. The long-used
     single ethnoracial group (n=3), or were inaccessible by our          Framingham risk factors, which were modelled using a largely
     research team (n=5).                                                 non-Hispanic White population, have recently been shown to
                                                                          inadequately capture risk in certain minority groups [65]. The
     The 10 articles selected for detailed analysis are summarized        STONE score to predict the likelihood of kidney stones in
     in Multimedia Appendices 1-3. Most articles were T2D-focused         patients with flank pain equates Black race with lower risk [66];
     (n=8), with the remaining articles focused on T1D (n=1) and          however, an external validation study found no significant
     gestational diabetes (n=1). The main report types were               association between non-Black race and increased risk of
     retrospective analyses of data pulled from electronic medical        developing kidney stones [67]. The Vaginal Birth after Cesarean
     records (n=5) or generated through randomized control trials         (VBAC) algorithm predicts a lower likelihood of successful
     (n=2). The reviewed articles reported ethnoracial data under         vaginal delivery in African American and Hispanic mothers
     various categories, including race (n=6), ethnicity (n=2),           having previously undergone cesarean section [68], while
     race/ethnicity (n=3), and percentage of Caucasian participants       ignoring other factors (eg, private insurance status, marital
     (n=1). Race was typically distributed between White (or              status) that have been significantly associated with VBAC
     Caucasian), Black (or African American), Asian, American             success [49]. A recent AI-based tool for classifying images of
     Indian, and Alaska Native. Ethnicity was generally reported as       skin cancer was reported to perform similarly to trained experts
     Hispanic and non-Hispanic. Among articles specifically               [69]; however, the training images were predominantly of
     reporting race, the average distribution was 69.5% White, 17.1%      light-skinned individuals, and performance was not assessed
     Black, and 3.7% Asian (Multimedia Appendix 1). The 2 articles        on those with darker skin [70]. These examples highlight the
     that specifically included ethnicity reported 7.2% and 21.3%         importance of effective ethnoracial considerations in the
     Hispanic patients (Multimedia Appendix 2) [53,61]. The average       development of clinical decision support tools.
     distribution in articles that merged race and ethnicity was 55.4%
     non-Hispanic White, 8.1% non-Hispanic Black, 19.9% Hispanic,         Despite the promise of AI, several factors predispose AI
     and 8.3% Asian (Multimedia Appendix 3). Only 2 articles              algorithms to negative biases. One limitation of AI models is
     reported inclusion of Native American participants [59,61]. The      the so-called distributional shift, where erroneous predictions
     sole non-American study was performed in the Netherlands and         result from a mismatch between the training population and the
     included 97.7% Caucasian participants [60].                          population on which the model is used. Such a mismatch can
                                                                          result from “bias in the training set, change over time, or use
     Several of the selected studies also included specific discussion    of the system in a different population” [50]. Essentially, the
     of ethnoracial themes. Rohan et al stated that their research was    robustness of AI algorithms is dependent upon the degree to
     limited by the homogeneity of their study population and that        which the training population represents the target population
     the generalizability of their findings should be further             [71]. In addition to the distributional shift phenomenon, the
     investigated [54]. Two more studies acknowledged that their          complexity and black-box nature of AI algorithms often
     study populations were mainly White [58,60], with one stating        obfuscates underlying errors or biases, specifically when
     that their predominantly White and female demographic was            compared to simpler rule-based systems [50]. The detection of
     “not uncommon in behavioral weight loss studies” [58]. Valdez        such biases in AI algorithms often requires careful consideration
     et al intentionally oversampled racial and ethnic minorities and     of model behavior in response to changing inputs [72]. In the
     identified very few ethnoracial differences in health information    case of ethnoracial data, the omission of such information could
     communication patterns [61]. McCoy et al noted that                  result in a distributional shift based on ethnicity, race, or both
     race/ethnicity did not contribute to their predictions of glycemic   in resultant models, which may be difficult for researchers to
     trajectory and proposed that ethnoracial disparities in glycemic     identify at the time of development.
     control may reflect differences in access to health care and
     medications [57].                                                    Given the clear ethnic and racial differences in diabetes
                                                                          biomarkers, prevalence, and outcomes [7-10,12,73], the
                                                                          inclusion of ethnoracial data is also likely to improve the
                                                                          accuracy of predictive models. The predictive value of race and

     http://www.jmir.org/2021/2/e22320/                                                            J Med Internet Res 2021 | vol. 23 | iss. 2 | e22320 | p. 3
                                                                                                                 (page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH                                                                                                   Pham et al

     ethnicity is well-documented in the literature, where they have     American participants were reported in just 2 studies [59,61],
     been shown to independently predict health decline for adults       despite making up an estimated 1.3% of the American
     living with diabetes [74,75]. The impact of specific risk factors   population [83] and being the ethnoracial group with the highest
     for T2D have even been shown to vary for both sex and race,         age-adjusted prevalence of diagnosed diabetes [9]. Poor
     with the most predictive factors being waist circumference in       Indigenous representation in health and governmental data sets
     Black men, 2-hour glucose from an oral glucose tolerance test       has been previously reported in the literature [85,86]. In Canada,
     in Black women, and fasting glucose in both White men and           where Indigenous peoples account for 4.9% of the population
     White women [76]. As a result of these associations between         [87] and are disproportionately affected by diabetes [88], failure
     diabetes outcomes and ethnoracial information, the consideration    to include Indigenous data when training diabetes AI models
     of ethnoracial data is likely to enhance both the accuracy and      could propagate existing issues of health inequity and structural
     generalizability of resultant AI-based diabetes tools.              racism in this population [89,90].
     In those articles we reviewed that did include ethnoracial          A Simple Screening Tool to Assess Ethnoracial Equity
     information, there was very little standardization in terms of      in Diabetes AI
     how these data were reported (eg, race, race and ethnicity,
                                                                         Detailed guidelines currently exist for the development of
     race/ethnicity). Race distinguishes individuals based on ancestry
                                                                         trustworthy and human-centric AI technologies [91]. However,
     and combinations of physical characteristics, whereas ethnicity
                                                                         we believe there is a need for simple tools to screen the ethnic
     focuses on behavior and culture in addition to physical features
                                                                         and racial generalizability of AI in health care. Based on the
     [77]. Inconsistent reporting of ethnic and racial information
                                                                         findings from our case study, we have developed a short
     hinders the ability to perform meta-analyses across multiple
                                                                         screening tool that researchers and clinicians may use to assess
     data sets and may limit ethnoracial equity in future AI
                                                                         ethnoracial equity in research describing AI-based diabetes
     applications [78]. In their writings on eliminating health
                                                                         interventions. The rationale and structure of this tool borrows
     disparities, Fremont and Lurie state that data pertaining to race
                                                                         from the Jadad scale [92], which was conceived by the founder
     and ethnicity are collected by a variety of sources, but “the
                                                                         of our research group over two decades ago and is widely used
     utility of these data is constrained by ongoing problems with
                                                                         to assess the methodological quality of a clinical trial [93,94].
     reliability, completeness, and lack of comparability across data
                                                                         We propose the following set of five questions to consider the
     sources” [79]. Though differences in the reporting of ethnoracial
                                                                         ethnoracial relevance of diabetes AI:
     data are expected across jurisdictions, we propose that authors
     attempt to report such data in a manner that is easily comparable   1.   Did the research explicitly describe the disease under study
     to locally available data. For example, the US census reports            (eg, T1D, T2D, both)? (1a) Did the research describe
     race and ethnicity separately, with ethnicity being used to              ethnoracial differences in disease prevalence, biomarkers,
     determine whether an individual is of “Hispanic origin or not”           and outcomes?
     and race being categorized as “White, Black or African              2.   Did the research clearly describe the sources of data used
     American, Asian, American Indian or Alaska Native, Native                in the training data set (eg, electronic medical record,
     Hawaiian or Other Pacific Islander, and other” [80]. Kiran et al         administrative data repository, research registry)? (2a) Did
     recently assessed Canadian patient perspectives on routinely             the research describe ethnoracial limitations in the sources
     being asked about their race and ethnicity through a                     of data?
     sociodemographic questionnaire [81]. They found that patients       3.   Did the research explicitly report the ethnic and racial
     were not uncomfortable disclosing their race and ethnicity and           backgrounds of individuals in the training data set? (3a)
     intuitively understood how the data could be helpful for their           Are ethnic and racial backgrounds reported in a manner
     health care providers. Their work has subsequently informed              that is easily comparable to local census data?
     the collection of race-based data during the COVID-19               4.   Do the ethnic and racial distributions in the training data
     pandemic [82]. These are just two examples of standards that             set accurately represent the population on which the
     will allow for comparison with locally available data and in             algorithm will be used? (4a) Did the research articulate
     turn enable the assessment of ethnoracial generalizability and           limitations of the ethnic and racial distributions in the
     cultural competence in diabetes AI algorithms.                           training data set?
                                                                         5.   Did the research describe strategies to mitigate ethnoracial
     In considering the average race distribution of the reviewed
                                                                              bias in the algorithm?
     studies, the proportions for White (69.5%) and Asian race
     (3.7%) were slightly lower than those values reported in recent     Although we feel that our proposed tool will be helpful in
     US census data (76.3% and 5.9%, respectively). The opposite         assessing clinical AI algorithms generally, it will be particularly
     was true for the Black race, which accounted for 17.1% of study     important in the development of diabetes AI. We believe that
     participants but only 13.4% of US census participants. In those     these innovations will fail to serve the diabetes community if
     studies reporting race and ethnicity as a combined variable, the    they are not trained on ethnoracially diverse data. As AI-based
     average proportion of non-Hispanic whites (55.4%) was slightly      systems become integrated into important aspects of diabetes
     lower than the census value of 60.1% [83]. These findings likely    management, such ethnoracial inequities in model development
     follow from the high prevalence of diabetes in the non-Hispanic     could ultimately be dangerous for minority groups whose
     Black population, specifically when compared to the                 biomarkers and outcomes may differ from the general
     non-Hispanic White and Asian populations [84]. One                  population. In the Contreras and Vehi review, most studies
     particularly worrisome finding was that data from Native            focused on T2D self-management, clinical decision support,

     http://www.jmir.org/2021/2/e22320/                                                           J Med Internet Res 2021 | vol. 23 | iss. 2 | e22320 | p. 4
                                                                                                                (page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH                                                                                                      Pham et al

     and prediction tools. Each of these dimensions of diabetes care        disproportionately affect ethnoracial minorities. The Centers
     can be affected by ethnoracial factors. For example, adherence         for Disease Control and Prevention have already determined
     to T2D medications to achieve euglycemia is demonstrably               that individuals from Black and American Indian or Alaska
     driven by cultural beliefs, values, social factors, religion, health   Native communities have a rate of hospitalization or death from
     literacy, and language barriers [95,96]. Similar issues are likely     COVID-19 that is 5 times greater than that of their White
     to follow in the T1D space, where AI algorithms are currently          counterparts [108]. It stands to reason that the increased
     focused on automated insulin delivery systems but will likely          prevalence of both COVID-19 and diabetes in ethnoracial
     shift toward the above dimensions in the near future [63,97,98].       minority groups and the relationship between these two
                                                                            conditions require ethnoracial considerations in all aspects of
     Addressing ethnoracial bias in diabetes AI has been made even
                                                                            diabetes care.
     more critical by the coronavirus disease 2019 (COVID-19)
     pandemic [99]. There is growing evidence to support a                  At this unprecedented time in history, AI can either mitigate or
     “bidirectional relationship between COVID-19 and diabetes”             exacerbate disparities in health care. Future accounts of the
     [100]. Research suggests that diabetes is a risk factor for rapid      infancy of diabetes AI must reflect our early and decisive action
     progression and poor prognosis of COVID-19 [101,102].                  to confront ethnoracial inequities before they are coded into our
     New-onset diabetes is also being reported in previously healthy        systems and perpetuate the very biases we aim to eliminate [45].
     individuals diagnosed with COVID-19 [103-105], which may               If we take deliberate and meaningful steps now toward training
     reflect coronavirus-inflicted damage to insulin-producing cells        our algorithms to be ethnoracially inclusive, we can architect
     [106,107]. We are concerned by these findings from a health            innovations in diabetes care that are bound by the diverse fabric
     equity lens, given that COVID-19 has been found to                     of our society.

     Conflicts of Interest
     None declared.

     Multimedia Appendix 1
     Distribution of articles specifically reporting race.
     [DOCX File , 14 KB-Multimedia Appendix 1]

     Multimedia Appendix 2
     Distribution of articles specifically reporting ethnicity.
     [DOCX File , 13 KB-Multimedia Appendix 2]

     Multimedia Appendix 3
     Distribution for articles reporting race and ethnicity as a merged variable.
     [DOCX File , 14 KB-Multimedia Appendix 3]

     References
     1.      Piccolo RS, Subramanian S, Pearce N, Florez JC, McKinlay JB. Relative Contributions of Socioeconomic, Local
             Environmental, Psychosocial, Lifestyle/Behavioral, Biophysiological, and Ancestral Factors to Racial/Ethnic Disparities
             in Type 2 Diabetes. Diabetes Care 2016 Jul 21;39(7):1208-1217 [FREE Full text] [doi: 10.2337/dc15-2255] [Medline:
             27330127]
     2.      Immigration and Ethnocultural Diversity in Canada. 2013. URL: https://www12.statcan.gc.ca/nhs-enm/2011/as-sa/99-010-x/
             99-010-x2011001-eng.cfm [accessed 2021-01-16]
     3.      Williams DR, Priest N, Anderson NB. Understanding associations among race, socioeconomic status, and health: Patterns
             and prospects. Health Psychol 2016 May;35(4):407-411 [FREE Full text] [doi: 10.1037/hea0000242] [Medline: 27018733]
     4.      Isaranuwatchai W, Fazli GS, Bierman AS, Lipscombe LL, Mitsakakis N, Shah BR, et al. Universal Drug Coverage and
             Socioeconomic Disparities in Health Care Costs Among Persons With Diabetes. Diabetes Care 2020 Oct;43(9):2098-2105.
             [doi: 10.2337/dc19-1536] [Medline: 32641377]
     5.      Agardh E, Allebeck P, Hallqvist J, Moradi T, Sidorchuk A. Type 2 diabetes incidence and socio-economic position: a
             systematic review and meta-analysis. Int J Epidemiol 2011 Jul 19;40(3):804-818. [doi: 10.1093/ije/dyr029] [Medline:
             21335614]
     6.      Bird Y, Lemstra M, Rogers M, Moraros J. The relationship between socioeconomic status/income and prevalence of diabetes
             and associated conditions: A cross-sectional population-based study in Saskatchewan, Canada. Int J Equity Health 2015
             Oct 12;14(1):93 [FREE Full text] [doi: 10.1186/s12939-015-0237-0] [Medline: 26458543]
     7.      Carson AP, Muntner P, Selvin E, Carnethon MR, Li X, Gross MD, et al. Do glycemic marker levels vary by race? Differing
             results from a cross-sectional analysis of individuals with and without diagnosed diabetes. BMJ Open Diabetes Res Care
             2016 Jun 10;4(1):e000213 [FREE Full text] [doi: 10.1136/bmjdrc-2016-000213] [Medline: 27335652]

     http://www.jmir.org/2021/2/e22320/                                                              J Med Internet Res 2021 | vol. 23 | iss. 2 | e22320 | p. 5
                                                                                                                   (page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH                                                                                                    Pham et al

     8.      Cavagnolli G, Pimentel AL, Freitas PAC, Gross JL, Camargo JL. Effect of ethnicity on HbA1c levels in individuals without
             diabetes: Systematic review and meta-analysis. PLoS One 2017 Feb 13;12(2):e0171315 [FREE Full text] [doi:
             10.1371/journal.pone.0171315] [Medline: 28192447]
     9.      Spanakis EK, Golden SH. Race/ethnic difference in diabetes and diabetic complications. Curr Diab Rep 2013 Dec
             15;13(6):814-823 [FREE Full text] [doi: 10.1007/s11892-013-0421-9] [Medline: 24037313]
     10.     Lanting LC, Joung IM, Mackenbach JP, Lamberts SW, Bootsma AH. Ethnic differences in mortality, end-stage complications,
             and quality of care among diabetic patients: a review. Diabetes Care 2005 Oct 25;28(9):2280-2288. [doi:
             10.2337/diacare.28.9.2280] [Medline: 16123507]
     11.     McBean AM, Li S, Gilbertson DT, Collins AJ. Differences in diabetes prevalence, incidence, and mortality among the
             elderly of four racial/ethnic groups: whites, blacks, hispanics, and asians. Diabetes Care 2004 Oct 24;27(10):2317-2324.
             [doi: 10.2337/diacare.27.10.2317] [Medline: 15451894]
     12.     Golden S, Brown A, Cauley JA, Chin MH, Gary-Webb TL, Kim C, et al. Health disparities in endocrine disorders: biological,
             clinical, and nonclinical factors--an Endocrine Society scientific statement. J Clin Endocrinol Metab 2012
             Oct;97(9):E1579-E1639 [FREE Full text] [doi: 10.1210/jc.2012-2043] [Medline: 22730516]
     13.     Kahkoska AR, Shay CM, Crandell J, Dabelea D, Imperatore G, Lawrence JM, et al. Association of Race and Ethnicity
             With Glycemic Control and Hemoglobin A Levels in Youth With Type 1 Diabetes. JAMA Netw Open 2018 Oct
             07;1(5):e181851 [FREE Full text] [doi: 10.1001/jamanetworkopen.2018.1851] [Medline: 30370425]
     14.     Healthy People 2020: An opportunity to address societal determinants of health in the United States. In: Secretary's Advisory
             Committee on Health Promotion and Disease Prevention Objectives for 2020. Washington, DC: US Department of Health
             and Human Services; 2010.
     15.     Walker RJ, Strom Williams J, Egede LE. Influence of Race, Ethnicity and Social Determinants of Health on Diabetes
             Outcomes. Am J Med Sci 2016 May;351(4):366-373 [FREE Full text] [doi: 10.1016/j.amjms.2016.01.008] [Medline:
             27079342]
     16.     Caruso R, Magon A, Baroni I, Dellafiore F, Arrigoni C, Pittella F, et al. Health literacy in type 2 diabetes patients: a
             systematic review of systematic reviews. Acta Diabetol 2018 Jan 11;55(1):1-12. [doi: 10.1007/s00592-017-1071-1] [Medline:
             29129000]
     17.     Connolly V, Unwin N, Sherriff P, Bilous R, Kelly W. Diabetes prevalence and socioeconomic status: a population based
             study showing increased prevalence of type 2 diabetes mellitus in deprived areas. J Epidemiol Community Health 2000
             Mar;54(3):173-177 [FREE Full text] [doi: 10.1136/jech.54.3.173] [Medline: 10746110]
     18.     Elbein SC. Genetics factors contributing to type 2 diabetes across ethnicities. J Diabetes Sci Technol 2009 Jul 01;3(4):685-689
             [FREE Full text] [doi: 10.1177/193229680900300412] [Medline: 20144314]
     19.     Dinca-Panaitescu S, Dinca-Panaitescu M, Bryant T, Daiski I, Pilkington B, Raphael D. Diabetes prevalence and income:
             Results of the Canadian Community Health Survey. Health Policy 2011 Mar;99(2):116-123. [doi:
             10.1016/j.healthpol.2010.07.018] [Medline: 20724018]
     20.     Gallegos-Macias A, Macias SR, Kaufman E, Skipper B, Kalishman N. Relationship between glycemic control, ethnicity
             and socioeconomic status in Hispanic and white non-Hispanic youths with type 1 diabetes mellitus. Pediatr Diabetes 2003
             Mar;4(1):19-23. [doi: 10.1034/j.1399-5448.2003.00020.x] [Medline: 14655519]
     21.     Scott A, Chambers D, Goyder E, O'Cathain A. Socioeconomic inequalities in mortality, morbidity and diabetes management
             for adults with type 1 diabetes: A systematic review. PLoS One 2017 May 10;12(5):e0177210 [FREE Full text] [doi:
             10.1371/journal.pone.0177210] [Medline: 28489876]
     22.     Kavakiotis I, Tsave O, Salifoglou A, Maglaveras N, Vlahavas I, Chouvarda I. Machine Learning and Data Mining Methods
             in Diabetes Research. Comput Struct Biotechnol J 2017;15:104-116 [FREE Full text] [doi: 10.1016/j.csbj.2016.12.005]
             [Medline: 28138367]
     23.     Gunasekeran DV. Technology and chronic disease management. The Lancet Diabetes & Endocrinology 2018 Feb;6(2):91.
             [doi: 10.1016/s2213-8587(17)30441-2]
     24.     Kaplan A, Haenlein M. Siri, Siri, in my hand: Who’s the fairest in the land? On the interpretations, illustrations, and
             implications of artificial intelligence. Business Horizons 2019 Jan;62(1):15-25. [doi: 10.1016/j.bushor.2018.08.004]
     25.     Tang A, Tam R, Cadrin-Chênevert A, Guest W, Chong J, Barfett J, Canadian Association of Radiologists (CAR) Artificial
             Intelligence Working Group. Canadian Association of Radiologists White Paper on Artificial Intelligence in Radiology.
             Can Assoc Radiol J 2018 May 01;69(2):120-135 [FREE Full text] [doi: 10.1016/j.carj.2018.02.002] [Medline: 29655580]
     26.     Pesapane F, Codari M, Sardanelli F. Artificial intelligence in medical imaging: threat or opportunity? Radiologists again
             at the forefront of innovation in medicine. Eur Radiol Exp 2018 Oct 24;2(1):35 [FREE Full text] [doi:
             10.1186/s41747-018-0061-6] [Medline: 30353365]
     27.     Li L, Qin L, Xu Z, Yin Y, Wang X, Kong B, et al. Using Artificial Intelligence to Detect COVID-19 and Community-acquired
             Pneumonia Based on Pulmonary CT: Evaluation of the Diagnostic Accuracy. Radiology 2020 Aug 19;296(2):E65-E71
             [FREE Full text] [doi: 10.1148/radiol.2020200905] [Medline: 32191588]
     28.     Jerez-Aragonés JM, Gómez-Ruiz JA, Ramos-Jiménez G, Muñoz-Pérez J, Alba-Conejo E. A combined neural network and
             decision trees model for prognosis of breast cancer relapse. Artif Intell Med 2003 Jan;27(1):45-63. [doi:
             10.1016/s0933-3657(02)00086-6] [Medline: 12473391]

     http://www.jmir.org/2021/2/e22320/                                                            J Med Internet Res 2021 | vol. 23 | iss. 2 | e22320 | p. 6
                                                                                                                 (page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH                                                                                                  Pham et al

     29.     Liu X, Chen K, Wu T, Weidman D, Lure F, Li J. Use of multimodality imaging and artificial intelligence for diagnosis and
             prognosis of early stages of Alzheimer's disease. Transl Res 2018 Apr;194:56-67 [FREE Full text] [doi:
             10.1016/j.trsl.2018.01.001] [Medline: 29352978]
     30.     Hearn J, Ross HJ, Mueller B, Fan C, Crowdy E, Duhamel J, et al. Neural Networks for Prognostication of Patients With
             Heart Failure. Circ Heart Fail 2018 Aug;11(8):e005193. [doi: 10.1161/CIRCHEARTFAILURE.118.005193] [Medline:
             30354561]
     31.     Horng S, Sontag DA, Halpern Y, Jernite Y, Shapiro NI, Nathanson LA. Creating an automated trigger for sepsis clinical
             decision support at emergency department triage using machine learning. PLoS One 2017 Apr 6;12(4):e0174708 [FREE
             Full text] [doi: 10.1371/journal.pone.0174708] [Medline: 28384212]
     32.     Valdes G, Simone CB, Chen J, Lin A, Yom SS, Pattison AJ, et al. Clinical decision support of radiotherapy treatment
             planning: A data-driven machine learning strategy for patient-specific dosimetric decision making. Radiother Oncol 2017
             Dec;125(3):392-397. [doi: 10.1016/j.radonc.2017.10.014] [Medline: 29162279]
     33.     Peiffer-Smadja N, Rawson T, Ahmad R, Buchard A, Georgiou P, Lescure F, et al. Machine learning for clinical decision
             support in infectious diseases: a narrative review of current applications. Clin Microbiol Infect 2020 May;26(5):584-595.
             [doi: 10.1016/j.cmi.2019.09.009] [Medline: 31539636]
     34.     Plis K, Bunescu R, Marling C, Shubrook J, Schwartz F. A machine learning approach to predicting blood glucose levels
             for diabetes management. 2014 Presented at: Workshops at the Twenty-Eighth AAAI conference on artificial intelligence;
             July 27-31, 2014; Quebec City, QC.
     35.     Gyuk P, Vassányi I, Kósa I. Blood Glucose Level Prediction for Diabetics Based on Nutrition and Insulin Administration
             Logs Using Personalized Mathematical Models. J Healthc Eng 2019 Jan 10;2019:8605206-8605212 [FREE Full text] [doi:
             10.1155/2019/8605206] [Medline: 30774850]
     36.     Lagani V, Chiarugi F, Manousos D, Verma V, Fursse J, Marias K, et al. Realization of a service for the long-term risk
             assessment of diabetes-related complications. J Diabetes Complications 2015 Jul;29(5):691-698. [doi:
             10.1016/j.jdiacomp.2015.03.011] [Medline: 25953402]
     37.     Dagliati A, Marini S, Sacchi L, Cogni G, Teliti M, Tibollo V, et al. Machine Learning Methods to Predict Diabetes
             Complications. J Diabetes Sci Technol 2018 Mar 12;12(2):295-302 [FREE Full text] [doi: 10.1177/1932296817706375]
             [Medline: 28494618]
     38.     Fioravanti A, Fico G, Arredondo MT, Leuteritz JP. A mobile feedback system for integrated E-health platforms to improve
             self-care and compliance of diabetes mellitus patients. In: Conf Proc IEEE Eng Med Biol Soc. 2011 Presented at: 2011
             Annual International Conference of the IEEE Engineering in Medicine and Biology Society; August 30-September 3, 2011;
             Boston, MA p. 3550-3553. [doi: 10.1109/iembs.2011.6090591]
     39.     Caballero-Ruiz E, García-Sáez G, Rigla M, Villaplana M, Pons B, Hernando ME. A web-based clinical decision support
             system for gestational diabetes: Automatic diet prescription and detection of insulin needs. Int J Med Inform 2017
             Jun;102:35-49. [doi: 10.1016/j.ijmedinf.2017.02.014] [Medline: 28495347]
     40.     Sowjanya K, Singhal A, Choudhary C. MobDBTest: A machine learning based system for predicting diabetes risk using
             mobile devices. 2015 Presented at: 2015 IEEE International Advance Computing Conference (IACC); 2015; Banglore p.
             397-402. [doi: 10.1109/IADCC.2015.7154738]
     41.     Zou Q, Qu K, Luo Y, Yin D, Ju Y, Tang H. Predicting Diabetes Mellitus With Machine Learning Techniques. Front Genet
             2018 Nov 6;9:515 [FREE Full text] [doi: 10.3389/fgene.2018.00515] [Medline: 30459809]
     42.     Parikh RB, Teeple S, Navathe AS. Addressing Bias in Artificial Intelligence in Health Care. JAMA 2019 Dec 22;322(24):2377
             [FREE Full text] [doi: 10.1001/jama.2019.18058] [Medline: 31755905]
     43.     Geneviève LD, Martani A, Shaw D, Elger BS, Wangmo T. Structural racism in precision medicine: leaving no one behind.
             BMC Med Ethics 2020 Feb 19;21(1):17 [FREE Full text] [doi: 10.1186/s12910-020-0457-8] [Medline: 32075640]
     44.     Sultana K, Sheikh A. Most UK datasets of routinely collected health statistics fail to collect information on ethnicity and
             religion. J R Soc Med 2008 Oct;101(9):463-465 [FREE Full text] [doi: 10.1258/jrsm.2008.080007] [Medline: 18779248]
     45.     Vyas DA, Eisenstein LG, Jones DS. Hidden in Plain Sight - Reconsidering the Use of Race Correction in Clinical Algorithms.
             N Engl J Med 2020 Aug 27;383(9):874-882 [FREE Full text] [doi: 10.1056/NEJMms2004740] [Medline: 32853499]
     46.     Peterson PN, Rumsfeld JS, Liang L, Albert NM, Hernandez AF, Peterson ED, American Heart Association Get With the
             Guidelines-Heart Failure Program. A validated risk score for in-hospital mortality in patients with heart failure from the
             American Heart Association get with the guidelines program. Circ Cardiovasc Qual Outcomes 2010 Jan;3(1):25-32. [doi:
             10.1161/CIRCOUTCOMES.109.854877] [Medline: 20123668]
     47.     Eneanya ND, Yang W, Reese PP. Reconsidering the Consequences of Using Race to Estimate Kidney Function. JAMA
             2019 Jul 09;322(2):113-114. [doi: 10.1001/jama.2019.5774] [Medline: 31169890]
     48.     Shahian DM, Jacobs JP, Badhwar V, Kurlansky PA, Furnary AP, Cleveland JC, et al. The Society of Thoracic Surgeons
             2018 Adult Cardiac Surgery Risk Models: Part 1-Background, Design Considerations, and Model Development. Ann
             Thorac Surg 2018 May;105(5):1411-1418. [doi: 10.1016/j.athoracsur.2018.03.002] [Medline: 29577925]
     49.     Landon M, Leindecker S, Spong CY, Hauth JC, Bloom S, Varner MW, National Institute of Child Health and Human
             Development Maternal-Fetal Medicine Units Network. The MFMU Cesarean Registry: factors affecting the success of trial

     http://www.jmir.org/2021/2/e22320/                                                          J Med Internet Res 2021 | vol. 23 | iss. 2 | e22320 | p. 7
                                                                                                               (page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH                                                                                                 Pham et al

             of labor after previous cesarean delivery. Am J Obstet Gynecol 2005 Oct;193(3 Pt 2):1016-1023. [doi:
             10.1016/j.ajog.2005.05.066] [Medline: 16157104]
     50.     Challen R, Denny J, Pitt M, Gompels L, Edwards T, Tsaneva-Atanasova K. Artificial intelligence, bias and clinical safety.
             BMJ Qual Saf 2019 Mar 12;28(3):231-237 [FREE Full text] [doi: 10.1136/bmjqs-2018-008370] [Medline: 30636200]
     51.     Centre for Global eHealth Innovation. URL: https://ehealthinnovation.org/ [accessed 2021-01-31]
     52.     Contreras I, Vehi J. Artificial Intelligence for Diabetes Management and Decision Support: Literature Review. J Med
             Internet Res 2018 May 30;20(5):e10775 [FREE Full text] [doi: 10.2196/10775] [Medline: 29848472]
     53.     Nunes AP, Yang J, Radican L, Engel SS, Kurtyka K, Tunceli K, et al. Assessing occurrence of hypoglycemia and its severity
             from electronic health records of patients with type 2 diabetes mellitus. Diabetes Res Clin Pract 2016 Dec;121:192-203.
             [doi: 10.1016/j.diabres.2016.09.012] [Medline: 27744128]
     54.     Rohan J, Delamater A, Pendley JS, Dolan L, Reeves G, Drotar D. Identification of self-management patterns in pediatric
             type 1 diabetes using cluster analysis. Pediatr Diabetes 2011 Dec;12(7):611-618 [FREE Full text] [doi:
             10.1111/j.1399-5448.2010.00752.x] [Medline: 21446925]
     55.     Anderson JP, Parikh JR, Shenfeld DK, Ivanov V, Marks C, Church BW, et al. Reverse Engineering and Evaluation of
             Prediction Models for Progression to Type 2 Diabetes: An Application of Machine Learning Using Electronic Health
             Records. J Diabetes Sci Technol 2015 Dec 20;10(1):6-18 [FREE Full text] [doi: 10.1177/1932296815620200] [Medline:
             26685993]
     56.     DuBrava S, Mardekian J, Sadosky A, Bienen EJ, Parsons B, Hopps M, et al. Using Random Forest Models to Identify
             Correlates of a Diabetic Peripheral Neuropathy Diagnosis from Electronic Health Record Data. Pain Med 2017 Jan
             01;18(1):107-115. [doi: 10.1093/pm/pnw096] [Medline: 27252307]
     57.     McCoy RG, Ngufor C, Van Houten HK, Caffo B, Shah ND. Trajectories of Glycemic Change in a National Cohort of
             Adults With Previously Controlled Type 2 Diabetes. Med Care 2017 Nov;55(11):956-964 [FREE Full text] [doi:
             10.1097/MLR.0000000000000807] [Medline: 28922296]
     58.     Everett E, Kane B, Yoo A, Dobs A, Mathioudakis N. A Novel Approach for Fully Automated, Personalized Health Coaching
             for Adults with Prediabetes: Pilot Clinical Trial. J Med Internet Res 2018 Feb 27;20(2):e72 [FREE Full text] [doi:
             10.2196/jmir.9723] [Medline: 29487046]
     59.     Hazlehurst BL, Lawrence JM, Donahoo WT, Sherwood NE, Kurtz SE, Xu S, et al. Automating assessment of lifestyle
             counseling in electronic health records. Am J Prev Med 2014 May;46(5):457-464 [FREE Full text] [doi:
             10.1016/j.amepre.2014.01.001] [Medline: 24745635]
     60.     Cleveringa FG, Welsing PM, van den Donk M, Gorter KJ, Niessen LW, Rutten GE, et al. Cost-effectiveness of the diabetes
             care protocol, a multifaceted computerized decision support diabetes management intervention that reduces cardiovascular
             risk. Diabetes Care 2010 Mar 23;33(2):258-263 [FREE Full text] [doi: 10.2337/dc09-1232] [Medline: 19933991]
     61.     Valdez R, Guterbock TM, Fitzgibbon K, Williams IC, Wellbeloved-Stone CA, Bears JE, et al. From loquacious to reticent:
             understanding patient health information communication to guide consumer health IT design. J Am Med Inform Assoc
             2017 Jul 01;24(4):680-696 [FREE Full text] [doi: 10.1093/jamia/ocw155] [Medline: 28069667]
     62.     Lo-Ciganic W, Donohue JM, Thorpe JM, Perera S, Thorpe CT, Marcum ZA, et al. Using machine learning to examine
             medication adherence thresholds and risk of hospitalization. Med Care 2015 Aug;53(8):720-728 [FREE Full text] [doi:
             10.1097/MLR.0000000000000394] [Medline: 26147866]
     63.     Dankwa-Mullan I, Rivo M, Sepulveda M, Park Y, Snowdon J, Rhee K. Transforming Diabetes Care Through Artificial
             Intelligence: The Future Is Here. Popul Health Manag 2019 Jun;22(3):229-242 [FREE Full text] [doi: 10.1089/pop.2018.0129]
             [Medline: 30256722]
     64.     Singla R, Singla A, Gupta Y, Kalra S. Artificial Intelligence/Machine Learning in Diabetes Care. Indian J Endocrinol Metab
             2019;23(4):495-497 [FREE Full text] [doi: 10.4103/ijem.IJEM_228_19] [Medline: 31741913]
     65.     Gijsberts CM, Groenewegen KA, Hoefer IE, Eijkemans MJC, Asselbergs FW, Anderson TJ, et al. Race/Ethnic Differences
             in the Associations of the Framingham Risk Factors with Carotid IMT and Cardiovascular Events. PLoS One 2015 Jul
             2;10(7):e0132321 [FREE Full text] [doi: 10.1371/journal.pone.0132321] [Medline: 26134404]
     66.     Moore C, Bomann S, Daniels B, Luty S, Molinaro A, Singh D, et al. Derivation and validation of a clinical prediction rule
             for uncomplicated ureteral stone--the STONE score: retrospective and prospective observational cohort studies. BMJ 2014
             Mar 26;348:g2191 [FREE Full text] [doi: 10.1136/bmj.g2191] [Medline: 24671981]
     67.     Wang RC, Rodriguez RM, Moghadassi M, Noble V, Bailitz J, Mallin M, et al. External Validation of the STONE Score,
             a Clinical Prediction Rule for Ureteral Stone: An Observational Multi-institutional Study. Ann Emerg Med 2016
             May;67(4):423-432.e2 [FREE Full text] [doi: 10.1016/j.annemergmed.2015.08.019] [Medline: 26440490]
     68.     Grobman W, Lai Y, Landon MB, Spong CY, Leveno KJ, Rouse DJ, National Institute of Child Health and Human
             Development (NICHD) Maternal-Fetal Medicine Units Network (MFMU). Development of a nomogram for prediction of
             vaginal birth after cesarean delivery. Obstet Gynecol 2007 May;109(4):806-812. [doi:
             10.1097/01.AOG.0000259312.36053.02] [Medline: 17400840]
     69.     Esteva A, Kuprel B, Novoa RA, Ko J, Swetter SM, Blau HM, et al. Dermatologist-level classification of skin cancer with
             deep neural networks. Nature 2017 Feb 02;542(7639):115-118. [doi: 10.1038/nature21056] [Medline: 28117445]

     http://www.jmir.org/2021/2/e22320/                                                         J Med Internet Res 2021 | vol. 23 | iss. 2 | e22320 | p. 8
                                                                                                              (page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH                                                                                                  Pham et al

     70.     Zou J, Schiebinger L. AI can be sexist and racist - it's time to make it fair. Nature 2018 Jul 18;559(7714):324-326. [doi:
             10.1038/d41586-018-05707-8] [Medline: 30018439]
     71.     Amodei D, Olah C, Steinhardt J, Christiano P, Schulman J, Mane D. Concrete Problems in AI Safety. arXiv. Preprint posted
             online on July 25, 2016.
     72.     Adler P, Falk C, Friedler SA, Nix T, Rybeck G, Scheidegger C, et al. Auditing black-box models for indirect influence.
             Knowl Inf Syst 2017 Oct 25;54(1):95-122. [doi: 10.1007/s10115-017-1116-3]
     73.     Brancati FL, Whelton PK, Kuller LH, Klag MJ. Diabetes mellitus, race, and socioeconomic status. A population-based
             study. Ann Epidemiol 1996 Jan;6(1):67-73. [doi: 10.1016/1047-2797(95)00095-x] [Medline: 8680628]
     74.     DeFronzo RA, Ferrannini E, Groop L, Henry RR, Herman WH, Holst JJ, et al. Type 2 diabetes mellitus. Nat Rev Dis
             Primers 2015 Jul 23;1:15019. [doi: 10.1038/nrdp.2015.19] [Medline: 27189025]
     75.     Nicklett EJ. Socioeconomic status and race/ethnicity independently predict health decline among older diabetics. BMC
             Public Health 2011 Oct 02;11(1):684 [FREE Full text] [doi: 10.1186/1471-2458-11-684] [Medline: 21888645]
     76.     Guo F, Garvey WT. Development of a Weighted Cardiometabolic Disease Staging (CMDS) System for the Prediction of
             Future Diabetes. J Clin Endocrinol Metab 2015 Oct;100(10):3871-3877 [FREE Full text] [doi: 10.1210/jc.2015-2691]
             [Medline: 26241327]
     77.     Edwards C, Fillingim RB, Keefe F. Race, ethnicity and pain. Pain 2001 Dec;94(2):133-137. [doi:
             10.1016/S0304-3959(01)00408-0] [Medline: 11690726]
     78.     Hernandez-Boussard T, Bozkurt S, Ioannidis JPA, Shah NH. MINIMAR (MINimum Information for Medical AI Reporting):
             Developing reporting standards for artificial intelligence in health care. J Am Med Inform Assoc 2020 Dec
             09;27(12):2011-2015 [FREE Full text] [doi: 10.1093/jamia/ocaa088] [Medline: 32594179]
     79.     National Research Council (US) Panel on DHHS Collection of Race and Ethnic Data, Ver Ploeg M, Perrin E. The Role of
             Racial and Ethnic Data Collection in Eliminating Disparities in Health Care. In: Eliminating Health Disparities: Measurement
             and Data Needs. Washington, DC: National Academies Press (US); 2004.
     80.     Jones NA, U.S. Census Bureau. Update on the U.S. Census Bureau's Race and Ethnic Research for the 2020 Census. URL:
             https://www.census.gov/content/dam/Census/newsroom/press-kits/2014/article_race_ethnic_research_2020census_jones.
             pdf [accessed 2021-01-16]
     81.     Kiran T, Sandhu P, Aratangy T, Devotta K, Lofters A, Pinto AD. Patient perspectives on routinely being asked about their
             race and ethnicity: Qualitative study in primary care. Can Fam Physician 2019 Aug;65(8):e363-e369 [FREE Full text]
             [Medline: 31413042]
     82.     Pinto AD, Hapsari A. Collecting data on race during the COVID-19 pandemic to identify inequities.: Upstream Lab URL:
             https://upstreamlab.files.wordpress.com/2020/04/
             upstream-lab-collecting-data-on-race-during-covid-brief-final-for-circulation-1.pdf [accessed 2021-01-16]
     83.     United States Census Bureau QuickFacts. URL: https://www.census.gov/quickfacts/fact/table/US/PST045219 [accessed
             2021-01-16]
     84.     Cheng YJ, Kanaya AM, Araneta MRG, Saydah SH, Kahn HS, Gregg EW, et al. Prevalence of Diabetes by Race and
             Ethnicity in the United States, 2011-2016. JAMA 2019 Dec 24;322(24):2389-2398 [FREE Full text] [doi:
             10.1001/jama.2019.19365] [Medline: 31860047]
     85.     Fremantle E, Zurynski YA, Mahajan D, D'Antoine H, Elliott EJ. Indigenous child health: urgent need for improved data
             to underpin better health outcomes. Med J Aust 2008 May 19;188(10):588-591. [doi: 10.5694/j.1326-5377.2008.tb01797.x]
             [Medline: 18484933]
     86.     Trevethan S. Strengthening the Availability of First Nations Data.: Indigenous Services Canada & The Assembly of First
             Nations; 2019. URL: https://www.afn.ca/wp-content/uploads/2019/05/
             NCR-11176060-v1-STRENGTHENING_THE_AVAILABILITY_OF_FIRST_NATIONS_DATA-MAR_25_2019-FINAL_E.
             pdf [accessed 2021-01-16]
     87.     Aboriginal peoples in Canada: Key results from the 2016 Census.: Government of Canada, Statistics Canada; 2017. URL:
             https://www150.statcan.gc.ca/n1/daily-quotidien/171025/dq171025a-eng.htm [accessed 2021-01-31]
     88.     Dyck R, Osgood N, Lin TH, Gao A, Stang MR. Epidemiology of diabetes mellitus among First Nations and non-First
             Nations adults. CMAJ 2010 Mar 23;182(3):249-256 [FREE Full text] [doi: 10.1503/cmaj.090846] [Medline: 20083562]
     89.     Greenwood M, de Leeuw S, Lindsay N. Challenges in health equity for Indigenous peoples in Canada. Lancet 2018 Apr
             28;391(10131):1645-1648. [doi: 10.1016/S0140-6736(18)30177-6] [Medline: 29483024]
     90.     Paradies Y. Colonisation, racism and indigenous health. J Pop Research 2016 Feb 18;33(1):83-96. [doi:
             10.1007/s12546-016-9159-y]
     91.     Lemonne E. Ethics Guidelines for Trustworthy AI.: FUTURIUM - European Commission; 2018. URL: https://ec.europa.eu/
             futurium/en/ai-alliance-consultation [accessed 2021-01-31]
     92.     Jadad AR, Moore R, Carroll D, Jenkinson C, Reynolds DM, Gavaghan DJ, et al. Assessing the quality of reports of
             randomized clinical trials: is blinding necessary? Control Clin Trials 1996 Mar;17(1):1-12. [doi:
             10.1016/0197-2456(95)00134-4] [Medline: 8721797]
     93.     Olivo S, Macedo LG, Gadotti IC, Fuentes J, Stanton T, Magee DJ. Scales to assess the quality of randomized controlled
             trials: a systematic review. Phys Ther 2008 Mar;88(2):156-175. [doi: 10.2522/ptj.20070147] [Medline: 18073267]

     http://www.jmir.org/2021/2/e22320/                                                          J Med Internet Res 2021 | vol. 23 | iss. 2 | e22320 | p. 9
                                                                                                               (page number not for citation purposes)
XSL• FO
RenderX
JOURNAL OF MEDICAL INTERNET RESEARCH                                                                                                           Pham et al

     94.     Halpern SH, Douglas MJ, editors. Appendix: Jadad Scale for Reporting Randomized Controlled Trials. In: Evidence-based
             Obstetric Anesthesia. Hoboken, NJ: Blackwell Publishing Ltd; Jun 2005:237-238.
     95.     Riddle MC. The underuse of insulin therapy in North America. Diabetes Metab Res Rev 2002 Sep;18 Suppl 3(S3):S42-S49.
             [doi: 10.1002/dmrr.277] [Medline: 12324985]
     96.     Rebolledo JA, Arellano R. Cultural Differences and Considerations When Initiating Insulin. Diabetes Spectr 2016 Aug
             15;29(3):185-190 [FREE Full text] [doi: 10.2337/diaspect.29.3.185] [Medline: 27574375]
     97.     Tyler NS, Jacobs PG. Artificial Intelligence in Decision Support Systems for Type 1 Diabetes. Sensors (Basel) 2020 Jul
             05;20(11):3214 [FREE Full text] [doi: 10.3390/s20113214] [Medline: 32517068]
     98.     Chaki J, Thillai Ganesh S, Cidham S, Ananda Theertan S. Machine learning and artificial intelligence based Diabetes
             Mellitus detection and self-management: A systematic review. Journal of King Saud University - Computer and Information
             Sciences 2020 Jul [FREE Full text] [doi: 10.1016/j.jksuci.2020.06.013]
     99.     Gamble A, Pham Q, Goyal S, Cafazzo JA. The Challenges of COVID-19 for People Living With Diabetes: Considerations
             for Digital Health. JMIR Diabetes 2020 May 15;5(2):e19581 [FREE Full text] [doi: 10.2196/19581] [Medline: 32392473]
     100.    Rubino F, Amiel SA, Zimmet P, Alberti G, Bornstein S, Eckel RH, et al. New-Onset Diabetes in Covid-19. N Engl J Med
             2020 Aug 20;383(8):789-790 [FREE Full text] [doi: 10.1056/NEJMc2018688] [Medline: 32530585]
     101.    Fang L, Karakiulakis G, Roth M. Are patients with hypertension and diabetes mellitus at increased risk for COVID-19
             infection? Lancet Respir Med 2020 Apr;8(4):e21 [FREE Full text] [doi: 10.1016/S2213-2600(20)30116-8] [Medline:
             32171062]
     102.    Guo W, Li M, Dong Y, Zhou H, Zhang Z, Tian C, et al. Diabetes is a risk factor for the progression and prognosis of
             COVID-19. Diabetes Metab Res Rev 2020 Mar 31:e3319 [FREE Full text] [doi: 10.1002/dmrr.3319] [Medline: 32233013]
     103.    Chee YJ, Ng SJH, Yeoh E. Diabetic ketoacidosis precipitated by Covid-19 in a patient with newly diagnosed diabetes
             mellitus. Diabetes Res Clin Pract 2020 Jul;164:108166 [FREE Full text] [doi: 10.1016/j.diabres.2020.108166] [Medline:
             32339533]
     104.    Marchand L, Pecquet M, Luyton C. Type 1 diabetes onset triggered by COVID-19. Acta Diabetol 2020 Oct
             11;57(10):1265-1266 [FREE Full text] [doi: 10.1007/s00592-020-01570-0] [Medline: 32653960]
     105.    Wang S, Ma P, Zhang S, Song S, Wang Z, Ma Y, et al. Fasting blood glucose at admission is an independent predictor for
             28-day mortality in patients with COVID-19 without previous diagnosis of diabetes: a multi-centre retrospective study.
             Diabetologia 2020 Oct 10;63(10):2102-2111 [FREE Full text] [doi: 10.1007/s00125-020-05209-1] [Medline: 32647915]
     106.    Yang J, Lin S, Ji X, Guo L. Binding of SARS coronavirus to its receptor damages islets and causes acute diabetes. Acta
             Diabetol 2010 Oct 31;47(3):193-199 [FREE Full text] [doi: 10.1007/s00592-009-0109-4] [Medline: 19333547]
     107.    Singh AK, Gupta R, Ghosh A, Misra A. Diabetes in COVID-19: Prevalence, pathophysiology, prognosis and practical
             considerations. Diabetes Metab Syndr 2020 Jul;14(4):303-310 [FREE Full text] [doi: 10.1016/j.dsx.2020.04.004] [Medline:
             32298981]
     108.    CDC. Coronavirus Disease 2019 (COVID-19). Centers for Disease Control and Prevention (2020). URL: https://www.
             cdc.gov/coronavirus/2019-ncov/need-extra-precautions/racial-ethnic-minorities.html [accessed 2021-01-31]

     Abbreviations
               AI: artificial intelligence
               SES: socioeconomic status
               T1D: type 1 diabetes
               T2D: type 2 diabetes
               VBAC: Vaginal Birth after Cesarean

               Edited by G Eysenbach; submitted 14.08.20; peer-reviewed by D Gunasekeran, K Gajewska; comments to author 13.10.20; revised
               version received 02.11.20; accepted 16.01.21; published 10.02.21
               Please cite as:
               Pham Q, Gamble A, Hearn J, Cafazzo JA
               The Need for Ethnoracial Equity in Artificial Intelligence for Diabetes Management: Review and Recommendations
               J Med Internet Res 2021;23(2):e22320
               URL: http://www.jmir.org/2021/2/e22320/
               doi: 10.2196/22320
               PMID: 33565982

     ©Quynh Pham, Anissa Gamble, Jason Hearn, Joseph A Cafazzo. Originally published in the Journal of Medical Internet Research
     (http://www.jmir.org), 10.02.2021. This is an open-access article distributed under the terms of the Creative Commons Attribution

     http://www.jmir.org/2021/2/e22320/                                                                  J Med Internet Res 2021 | vol. 23 | iss. 2 | e22320 | p. 10
                                                                                                                        (page number not for citation purposes)
XSL• FO
RenderX
You can also read